from intensive to reflexive: the prosodic factor · from intensive to reflexive: the prosodic...

[in E. König & V. Gast (sld.) Reciprocals and Reflexives -- Cross-linguistic and theoretical explorations : 591-631. Berlin : Mouton-De Gruyter]

From intensive to reflexive: the prosodic factor

Anne Zribi-Hertz

1. Introduction1

In this paper I return to the interpretive contrast first noted in Zribi-Hertz (1980) and also discussed in Zribi-Hertz (1995, 2003), between the English examples in (1) and their French counterparts in (2):

(1) a. b. ? c.

Johnz is proud of him∗z/k . *I am proud of me.

Johnz is proud of himz self.

d. I am proud of myself. (2) a.

b. Jeanz est fier de luiz/k. Je suis fier de moi.

(= [1a]) (= [1b])

c. Jeanz est fier de luiz -meˆme. (= [1c]) d. Je suis fier de moi-meˆme. (= [1d])

English simplex pronouns of the him paradigm (henceforth, him) exhibit disjoint-reference effects in prepositional contexts such as (1a), while French pro-

nouns of the lui paradigm (henceforth, lui) allow for the coreferential reading in examples such as (2a). As argued in Zribi-Hertz (1980, 1995, 2003), the availability of the coreferential reading in French is crucially sensitive to the lexical

context, and more precisely, to whether or not the semantic relation expressed

1 My thanks to Joaquim Branda˜o de Carvalho, Daniel Bu¨ring, Volker Gast, Ekkehard Ko¨nig and Eric Reuland, for the feedback they sent my way while I was writing this paper. I also have a debt of gratitude towards many colleagues, friends, relatives and students who helped me throughout the years sort out the French, British, American and creole data: among them Nicolas Ruwet, Lelia Picabia, Karl Gadelii, Philip Miller, Herby Glaude, Ge rard, Isabelle and Thomas Zribi, Claudia Morrissey, Rita Planey, Bridget Conlon, Nick, Joyce and Beatriz Belfrage, and Moby Pomerance. The responsibility for the resulting description is of course, entirely my own.

596 Anne Zribi-Hertz

z+k/k

by the predicate and the preposition is other-directed (±od),2 i.e. favours or disfavours disjoint reference, as illustrated by (3) vs. (4):

(3) [-od] contexts: coreference felicitous a. Jeanz est fier/content de

luiz/k. ‘John is proud of/pleased with 3m.sg’

b. Jeanz doit penser a luiz/k ‘John must think of/about 3m.sg’

c. Jeanz a honte de luiz/k. ‘John is ashamed of 3m.sg’ lit.

d. Jeanz est inquiet pour luiz/k. ‘John is worried about 3m.sg’

e. Je suis content de moi. ‘I am pleased with 1sg’

(4) [+od] contexts: coreference infelicitous a. Jeanz est de´pendant/jaloux de

∗z/k ‘John is dependent on/jealous of 3m.sg’

b. Jeanz {tient/est attache´} a lui∗z/k. ‘John is attached to 3m.sg’

c. Jeanz a besoin de lui∗z/k. ‘John has need of 3m.sg’

d. Jeanz est utile pour lui∗z/k ‘John is useful for 3m.sg’

e. *Je suis de´pendant de moi. ‘I am dependent on 1sg’

The semantic property labelled [±od] may be assessed independently from pronoun anaphora. In the following English and French examples, the internal and external arguments of [−od] predicates may intersect in reference (licensing the inclusive reading transcribed as ‘z + k’), while [+od] predicates disallow referential intersection and thus force their arguments to be construed as referentially disjoint:

(5) [−od] contexts: referential intersection felicitous a. Chomskyz {is {proud of/ashamed of/worried about}/must think

of } the MIT linguistsz+k/k . b. Chomskyz {{est fier des/a honte des/est inquiet pour les}/doit pen-

ser aux} linguistesz+k/k du MIT. (6) [+od] contexts: referential intersection disallowed > DR effect

a. Chomskyz is {jealous of/dependent on/attached to} the MIT linguists ∗z+k/k .

b. Chomskyz {est jaloux des/est de´pendant des/tient aux} linguistes

∗ du MIT.

In English, although referential intersection is possible between the lexical arguments of [-od] predicates such as proud+of, coreference is impossible when

2 This term is borrowed from Ko¨nig and Vezzozi (2004).

lui

From intensive to reflexive: the prosodic factor 597

the internal argument is a pronoun, as in (1a). In French, on the other hand, coreference is available for lui wherever referential intersection is allowed for a lexical noun phrase.

I shall admit the well-supported diachronic assumption (Faltz 1985; Levin- son 1991; Keenan 2002; van Gelderen 1999; Ko nig and Siemund 1999, 2000a,b; Ko nig and Vezzosi 2004) that complex pronouns such as English himself and French lui-me me, which were labelled M-pronouns in Zribi-Hertz (1995), start out as intensified pronouns, in the sense of Ko nig (1991), Ko nig and Siemund (1999) and Siemund (2000), and may eventually develop into reflexivity markers. As complex words, M-pronouns are formed of a simplex pronoun minimally specified for person,3 and of a focus particle (self or meˆme) which Ko nig (1991) calls an “intensifier”.4 At word-level, the pronoun component of M-pronouns is unstressed, with word stress falling on the intensifier self or meˆme. Correl- atively, M-pronouns are banned from the ostensive use which would require focal accent5 on the pronoun itself (Zribi-Hertz 1995). As regards interpretation, intensifiers are characterized by Ko nig (1991), Ko nig and Siemund (1999, 2000a,b), Ko nig and Vezzosi (2004) and Siemund (2000) as a class of focus markers involving the selection of one or a subset of a given set of referents.6 Un- der Ko nig’s (1991) analysis, what semantically characterizes intensifiers among other focus markers is that they signal the selected referent as “central” (Ko nig 1991), as opposed to the other members of the set construed as “peripheral”. From a syntactic point of view, English self-pronouns (henceforth, himself7) used as intensifiers occur as noun-phrase adjuncts in such examples as (7):

3 English self-pronouns are morphologically specified for person, number and semantic gender in the 3rd-person. French me me-pronouns are similarly specified with the exception of soi-meˆme, discussed below, which is unspecified for gender and number.

4 Whether we should analyze English self-pronouns as possessive nominals (myself parallel to my book) or as adjunction structures (English me+self parallel to German Hans selbst) is an open issue which is not directly relevant for this study. In any case, French meme-pronouns clearly cannot have a possessive structure, since me me is an adjective.

5 I will be using two different terms, accent and stress, to refer to phrasal and word prosody, respectively.

6 This type of focalization corresponds to what Erteschik-Shir (1997) calls restrictive focus.

7 Within the running text, capitalized pronouns (him, himself, lui, lui-meme) denote paradigms (e.g., him stands for me, you, him, her, us, them). Within examples, however, capitals are used as in Bu ring (1997, 2005) to indicate the position of primary accent.


(7) John took possession of his new office. He opened all the cabinets and all

the drawers and found many interesting documents and carbon copies of letters sent to various people. In the top right-hand one was an envelope addressed to Johnz himself z.

Baker (1995), Ko nig and Siemund (1999, 2000a,b), Ko nig and Vezzosi (2004) and Siemund (2000) propose to analyze A-free8 occurrences of himself, as illustrated in (8a), as syntactically parallel to the case illustrated in (7), i.e. as noun-phrase adjuncts supported by a pronoun reduced under identity, as represented in (8b):

(8) [same context as (7)] a. In the top right-hand one was an envelope addressed to himself.

= b. In the top right-hand one was an envelope addressed to [ [øz] himself z].

As argued in some detail by Baker (1995), this analysis correctly predicts the semantic properties of A-free himself, which are thoroughly similar to those of adnominal himself.

Under this general analysis, whenever Modern English himself is not overtly adjoined to a noun phrase as in (7), it is either bound by an argument, i.e. A- bound, as in (1c,d), or A-free, as in (8a). In the former case it is assumed to occur in argument position, while in the latter case it is assumed to be adjoined to a covert pronoun.

Like English himself, French lui-meme is morphologically an intensified pronoun. It however appears that the distribution of lui-meme in Modern French is more restricted than that of English himself. One obvious difference between English and French which correlates with the himself/lui-meme distributional contrast is that in French, non-clitic pronouns (the ones which may support an intensifier and thus form M-pronouns) are only available in a subset of noun- phrase positions. In particular, accusative and dative pronominal arguments must be realized as clitics, and correlatively, cannot be realized as strong pronouns, as shown in (9):

8 The occurrences of himself which I call A-free (A for argument) are called locally free in Chomsky (1981) and Baker (1995), exempt anaphors in Pollard and Sag (1992) and logophors in Reinhart and Reuland (1993).


(9) lexical noun phrase

a. Jean voit Paul.

a’. clitic pronoun Jeanz le∗z/k voit.

nonclitic pronoun a”. *Jean voit lui.

‘John sees Paul.’ ‘John sees him.’ b. Jean parle a Paul. b’. Jeanz lui∗z/k parle. b”. *Jean parle a lui.

‘John is talking to Paul.’ ‘Jean is talking to him.’

The clitic pronouns in (9a’) and (9b’) exhibit regular disjoint-reference effects – they can never be A-bound. The so-called “voix pronominale” (‘pronominal voice’), an inflectional paradigm involving a special clitic (se in the third person) obligatorily co-indexed with the local subject, as well as special auxiliary selection (e tre, with all verbs), is the only available reflexive-marking strategy for accusative and dative arguments:9

(10) a. Jeanz sez/∗k voit. ‘John sees himself.’

a’. *Jean voit lui(-meˆme).

b. Jeanz sez/∗k parle. b’. *Jean parle a lui(-meˆme). ‘John is talking to himself.’

French clitic pronouns are affix-like elements (Kayne 1975) which are morpho-

logically attached to a verb or auxiliary (Miller 1992; Miller and Monachesi 2003). In declarative clauses, they surface as proclitics. They cannot bear focal accent,10 nor be conjoined or modified, nor support an intensive adjunct. In order

9 For a recent and enlightening analysis of the voix pronominale, cf. Reinhart and Siloni (2005).

10 In imperative clauses, French object clitics occur postverbally, hence may fall under the phrase-final accent. They however display properties which qualify them as clitics: morphological attachment to the verb, inability to be modified or conjoined, or to fall under narrow focus (cf. Miller 1992), cf.: (i) a. Parle-moi.

talk-1sg ‘Talk to me.’

b. *Parle-moi, pas (a ) lui. b’. Parle-moi a moi, pas a lui. talk-1sg.cl neg (to) 3m.sg talk-1sg.cl to 1sg neg to 3m.sg ‘Talk to me, not to him.’

c. *Parle-moi et lui. talk-1sg and 3m.sg

d. *Parle-moi seul. talk-1sg alone


for the referent of a clitic pronoun to be construed as focused, the clitic must be doubled by a strong simplex or complex pronoun in postverbal position, e.g.:11

(11) a. Ce this

film lz’ film 3m.sg.acc

ennuie bore.prs.3sb

{luiz

3m.sg (-meme)}.12

-int

‘This film bores {him/even himself}.’ b. Jean sez voit {luiz(-meme)}.

‘John sees (even) himself.’

The strong pronoun in such structures may be assumed to be a non-argument (cf. Kayne 2001) since it bears no features of its own – its features replicate those of the clitic, maybe with the exception of Case which, if “absorbed” by the clitic, should be unspecified on the strong pronoun. That the clitic-doubling strong pronoun should be deficient for Case is supported by the fact that it does not alternate with a lexical noun phrase, as witnessed by the contrast between (11a) and (12):

(12) *Ce this

film film

l’ 3m.sg.acc

ennuie bore.prs.3sg

Jean. John

11 This description is at odds with the one proposed in Cardinaletti and Starke (1999), who give such examples as (i) as well-formed with the pronoun elle construed under narrow focus:

(i) *J’ ai vu Marie puis j’ ai vu elle. 1sg have.prs.1sg seen Mary and then 1sg have.prs.1sg seen 3f.sg

I reject this example as ill-formed, and so do all my French-speaking consultants, however strongly the pronoun may be stressed. For (i) to become grammatical, a clitic (agreeing with the strong pronoun) must be inserted, even if the referent of the pronoun should be in sight of the speaker:

(ii) J’ ai vu Mariek puis je l’z ai 1sg have.prs.1sg seen Mary and then 1sg 3f.sg have.prs.1sg vue ellez. seen.f.sg 3f.sg ‘I saw Maryk and then I saw herz.’

12 Abbreviations used in glosses: acc = accusative; cl = clitic; dat = dative; def = definite article; dem = demonstrative; f = feminine (gender); inf = infinitive; impf = imperfect tense; int = intensifier; m = masculine (gender); neg = negation; nom = nominative; pl = plural; pp = past participle; prs = present tense; pst = past; sg = singular; sbjv = subjunctive mood; 1, 2, 3 = first, second, third person.


∗z/k

French pronouns of the lui paradigm may thus be analyzed as arguments whenever they alternate with lexical noun phrases (as in [2]), and as non-arguments (adjuncts) when they occur as clitic doublers, as in (11). Since accusative and dative pronouns are realized as clitics, non-clitic lui only occurs as an argument under a preposition, as exemplified in (2), (3) and (4).13

The English/French interpretive contrast illustrated above by (1a,b) vs. (2a,b) is, crucially, only observed in contexts involving lui occurring as an argument. Whenever arguments are spelt out as clitics in French, they exhibit the same disjoint-reference effect as their English translations, regardless of the semantic features of the predicate:

(13) [−od] predicates a. Chomskyz defended the MIT linguistsk/z+k . b. Chomskyz a de fendu les linguistes du MIT k/z+k. c. Chomskyz defended him∗z/k . d. Chomskyzl

(14) [+od] predicates ’a de fendu.

a. Chomskyz hates the MIT linguistsk/∗z+k . b. Chomskyz de teste les linguistesk/∗z+k du MIT. c. Chomskyz hates him∗z/k. d. Chomskyz le∗z/kde´teste.

In what follows, I will further explore the contrast between simplex non-clitic pronouns (English him and French lui) and their complex counterparts (English himself and French lui-meme). I will show that French lui-meme is globally more restricted in its distribution than English himself, both as an argument and as a non-argument. I will argue that the different prosodic properties of English pronouns and French non-clitic pronouns might have contributed to their different semantics in the two languages.

I will first review and discuss (Section 2) the assumptions put forward so far in the linguistic literature to account for the interpretive contrast between (1) and (2), and will conclude that none of them provides a complete or satisfactory

13 This rough description leaves aside nominative pronouns, which although prosodically weak and morphologically attached to the right-hand context, have been shown to behave like phrasal affixes, rather than word affixes (Kayne 1975; Miller 1992; Car- dinaletti and Starke 1999). This issue may be disregarded for the present discussion, which focuses on the development of reflexivity markers. The relevant distinction here is that between prosodically attached pronouns, which I call clitics, and prosodically unattached ones, which I call non-clitics.


account of the observed contrast. I will then compare the prosodic properties of himself and lui-meme, and will show that lui-meme is always construed under narrow focus, while English himself fails to be similarly restricted. I will then propose to relate this contrast to the different prosodic properties of English pronouns and French non-clitic pronouns: English pronouns are leaners, in the sense of Zwicky (1982), i.e. they undergo deaccenting, a property which is observed for both simplex him and complex himself, whereas French lui may only be deaccented if a preceding word is under narrow focus, and French lui-meme is never deaccented. I will argue that the prosodic deaccenting of him- type pronouns – a property common to English pronouns and French clitics – favours their topic-binding as opposed to their argument-binding, so that the prosodic weakness of him might have favoured the development of English himself as a reflexive marker. French lui, on the other hand, fails to exhibit a prosodic weakness which could favour topic-binding over argument-binding. Since its reflexive reading is not disfavoured by prosody, it may only be hindered by the [+od] semantic effect of the lexical context. As a result, A-bound lui- me me is but a special instance of narrow focalization, while A-bound himself is syntactically motivated by argument-binding itself. The inherent prosodic strength of French lui further accounts for the fact that lui-me me must bear strong accent correlating with narrow focus, whereas English himself is not similarly restricted.

2. Previous analyses of “Condition B violations”

2.1. Binding vs. coreference

In Zribi-Hertz (1980), followed by Bouchard (1984), the co-indexing of lui with the local subject in French (2a) is assumed to transcribe coreference – a special case of referential intersection – rather than binding. This assumption is supported by the parallel between pronoun coreference and referential intersection, illustrated above in (3)–(4) and (5b)–(6b).

This theory, however, fails to explain why English him cannot similarly corefer with an argument in [-od] contexts, i.e. why there is a semantic contrast between (1a) and (5a) in English. Moreover, this theory of French lui conflicts with the fact that when it is co-indexed with the local subject, as in (15a), this pronoun may be construed as a bound variable, exactly as English himself in (15b):


(15) a. Jeanz

John est be.prs.3sg

fier de proud.m.sg of

luiz, 3m.sg

et and

Paulk Paul

aussi (z/k ). too ‘John is proud of himself, and Paul (is) too.’

b. Johnz is proud of himself z, and so is Paulk (z/k ).

These examples both allow either the referential (‘z’) or the so-called “sloppy” (i.e. variable) reading for the reconstructed pronoun within the elliptical predicate. They thus fail to support the assumption that the semantic relation between the pronoun and its binder/antecedent is of a crucially different nature for En- glish reflexive himself and for reflexive-read lui in French.

2.2. LUI as a “fourth-type” expression

Reasoning within the Standard Binding Theory framework, Ronat (1982) proposes that French non-clitic pronouns form a “fourth type” of expressions – alongside anaphors, pronominals and r-expressions – which are ambiguous between anaphors and pronominals. Under this view, the interpretive contrast between English him and French lui is due to the fact that him is a pronominal constrained by Binding Condition B, whereas French lui is a fourth-type expression, which has no equivalent in English. Ronat assumes that among French pronouns (se excepted), only clitics qualify as pronominals with respect to the Binding Theory.

This theory brings out an important parallel between English him and French clitics, treating French lui as special. It however fails to explain why French non-clitic pronouns should be exempt from Condition B, why clitichood should lead to disjoint-reference effects, and why English him, which is not a clitic, should be interpreted like French clitics rather than like French non-clitics.

2.3. The Avoid Pronoun theory

Pica (1984, 1986) proposes to derive the contrast between English (1a, b) and French (2a, b) from a general economy principle, “Avoid Pronoun”, which states that pronominals should be avoided whenever possible, in particular when a more specialized strategy is available in the language to convey the intended reading.14

The Avoid Pronoun principle would thus predict that him must be avoided in (1a) under the ‘z’ reading because a more specialized form (himself) is available

14 Similar ideas are put forward (albeit not specifically applied to French-English comparison) by Edmonson and Plank (1978) and Levinson (1991).


here for the reflexive reading. A similar analysis may account for the regular “Condition B” behaviour of French le-type clitics (cf. [9a’,b’]), which would compete with the voix pronominale under the reflexive reading (cf. [10]). Under Pica’s theory, the interpretive contrast between English him (1a) and French lui (2a) must be correlated with the fact that him in (1a) competes with a specialized reflexive-marking device (himself), whereas no specialized reflexive-marking device is available in French in positions calling for non-clitic pronouns.

This theory correctly emphasizes the crucial relevance of grammatical economy for the distribution and interpretion of linguistic expressions: thus, the distribution and interpretation of English him or French le are dependent on the distribution and interpretation of English himself and French se. How- ever, grammatical economy does not suffice to account for the observed data. As regards French clitics, all of them (se excepted) exhibit disjoint-reference effects, although some of them do not compete with the voix pronominale. For instance, the verb penser ‘think’ selects a locative complement introduced by the preposition a , which pronominalizes either as a +lui (16b) or as y, the locative clitic (16c); the voix pronominale is unavailable here (cf. [16d], for it is a priori restricted to dative and accusative arguments; nevertheless, the clitic pronoun y exhibits a disjoint-reference effect in [16c], while non-clitic lui may corefer with Jean in [16b]):

(16) a. Jean John

pense think.prs.3sg

a about

Paul. Paul

‘John is thinking about Paul.’ b. Jeanz

John pense think.prs.3sg

a about

luiz/k. 3m.sg

‘John is thinking about him(self).’ c. Jeanzy∗z/k pense.

‘John is thinking about {him/her/them/it}.’ d. *Jean se pense.

These data suggest that there might be some correlation between clitichood and disjoint-reference effects regardless of the availability of the voix pronominale to convey the reflexive reading.

Moreover, Pica’s theory is based on the common belief that French has only one morphological device specialized in reflexivity-marking: the voix pronominale, which is only licensed if the internal argument is accusative or dative. It follows that whenever these conditions are not met, no specialized reflexive- marking device is available, hence ordinary simplex pronouns (lui) take over


z/k

the reflexive reading. However, the voix pronominale can hardly be claimed to be “specialized in reflexive marking”, since it may also correlate with reciprocal, mediopassive and anticausative readings. Neither can it be claimed that French lui does not compete with a morphology specialized in reflexive-marking, since lui-me me triggers a reflexive reading in such cases as (17b):

(17) a. Jeanz

John est be.prs.3sg

atrocement horribly

jaloux jealous

de lui . ∗

of 3m.sg lit. ‘Johnz is horribly jealous of himz.’

b. Jeanz

John est be.prs.3sg

atrocement horribly

jaloux jealous

de luiz-meˆme. of 3m.sg-meˆme

‘John is horribly jealous of himself.’

While the [+od] predicate jaloux ‘jealous’ triggers a disjoint reading for lui in (17a), the disjoint-reference effect disappears in (17b) when lui-me me occurs. Since lui-me me makes the reflexive reading available in contexts where it is disallowed for simplex lui, we are entitled to claim that lui-me me qualifies as a morphological strategy “specialized in reflexive-marking”. Under this assumption, the Avoid Pronoun theory incorrectly predicts that lui and lui-me me should generally exhibit complementary interpretations in argument positions.

2.4. The Inalienable Pronoun theory

J. Rooryck and G. Van den Wyngaerd15 propose to analyze the interpretive contrast between English him (1a) and French lui (2a) on a par with (18):

(18) a. Jean John

a have.prs.3sg

leve raise.pp

le def.m.sg

doigt. finger

(i) ‘John raised the finger.’ (ii) ‘Johnz raised hisz finger.’ b. J’

1sg ai have.prs.1sg

leve raise.pp

le def.m.sg

doigt. finger

(i) ‘I raised the finger.’ (ii) ‘I raised my finger.’ c. John raised the finger. d. I raised the finger.

The French sentence in (18a) allows either for an alienable reading of the definite object le doigt, or for its inalienable reading. In English, (18c), the literal

15 In a talk on ‘Anaphora, identity and dissociation’, presented at the round-table on reflexives, Universite Paris-7/Leiden University, 1999.


l

translation of (18a), is only open to the alienable reading. Rooryck and Van den Wyngaerd’s assumption is that the semantic mechanism which provides the inalienable reading of the definite object in (18a) is the same as that which provides the reflexive interpretation of lui in (2a). Evidence in support of this idea is that the contrast between French (18a) and English (18c) is crucially linked to the definite article, as witnessed by (19), where in the presence of the indefinite article, English and French no longer contrast:

(19) a. John raised a finger. b. Jean

John a have.prs.3sg

leve raise.pp

un a.m.sg

doigt. finger

[± alienable finger in both examples]

Rooryck and Van den Wyngaerd’s theory is grounded in the assumption that English and French pronouns (e.g. him, lui) and definite articles form a single syntactic category (cf. Postal 1969; Emonds 1985) – a claim quite consistent with diachronic data.

This theory however runs into at least three problems. First, it does not explain why English and French pronoun-articles should have different semantic behaviours in (1a) and (2a). Second, the morphological unity of definite articles and pronouns only obtains in the third person, while the English-French contrasts illustrated in (1)–(2) and (18) are observed regardless of person. Third, since clitic le and non-clitic lui are both historically derived, in French, from the same paradigm of demonstratives (Latin ille), the theory fails to predict the sharp contrast between clitic and non-clitic pronouns with respect to disjoint- reference effects (cf. also [16b,c]):

(20) a. Jeanz ∗z/k’ a photographie´. John 3sg.acc have.prs.3sg photograph.pp ‘John photographed him(*self).’

b. Jeanz John

a have.prs.3sg

pris take.pp

une a.f.sg

photo de picture of

luiz/k. 3m.sg

‘John took a picture of him(self).’ (21) a.

Jeanz

John

∗z/k 3sg.dat

fait do.prs.3sg

confiance. trust

‘John trusts him(*self)/her.’ b. Jeanz

John a have.prs.3sg

confiance trust

en luiz/k . in 3m.sg

lit. ‘John has trust in him(self).’

lui


2.5. The Case-and-Chain Theory

Reinhart and Reuland (1993), Reuland and Reinhart (1995) and Reuland (1999, 2001, this volume) propose a theory of argument chains (A-chains) which may contribute to account for the interpretive contrast between English him and French lui in (1a) vs. (2a). An A-chain is defined as a sequence of co-indexation which is headed by an argument position (A-position) and satisfies antecedent government: each co-indexed link of the chain is c-commanded by the one above it. Thus, in such examples as (1) and (2), the pronoun and its antecedent form an A-chain. A-chains are assumed to abide by the general condition reproduced in (22):

(22) General condition on A-chains (Reuland 1999: 23) A maximal A-chain (α 1..............α n) contains exactly one link (α 1) which is fully specified for φ -features.

This means that the bound anaphor which stands at the foot of the chain must be deficient as to its content. The authors formalize this property in terms of φ-features: bound anaphors must be φ-deficient in one way or another. This condition does not obtain for English him in (1a), which is specified for number and semantic gender and is further assumed by Reinhart and Reuland to be specified for structural Case. Correlatively, him cannot form a chain with John in (1a), so that a disjoint-reference effect is observed, while himself can form a chain with John in (1c) because it is deficient for Case. This idea finds support in the fact that like many anaphors, himself fails to have a nominative form (Reinhart and Reuland 1993). Under this theory, the availability of the coreferential reading for lui in (2a) may be derived from the assumption that lui in (2a), unlike him in (1a), is φ-deficient. Since him and lui are both overtly specified for person, gender and number, the feature deficiency of lui must involve some abstract syntactic property. Reuland (this volume) thus assumes that unlike English him in (1a), French lui in (2a) is not specified for structural Case, but only for oblique Case, taken as a deficient value for Case. It follows that French lui, unlike English him, may stand at the foot of an A-chain without violating (22). English him, on the other hand, is analyzed by Reuland (1999, this volume) as specified either for structural Case or for oblique Case. Thus in (23a), him is assumed to receive structural Case, whence the disjoint-reference effect, while in (23b) it is assumed to receive oblique Case and may therefore form a chain with John:

(23) a. Johnz is ashamed of him∗z/k . b. Johnz looked behind himz/k.


This theory could find historical support in the fact that Old English drew a morphological distinction between accusative hine and oblique him (van Gelderen 1999). Under Reuland’s theory, Modern-English him has hung on to the accusative/oblique distinction in syntax, although it has neutralized it in morphology. French lui, on the other hand, has always been an oblique pronoun in all its occurrences.

The Case-and-Chain theory relates the interpretive properties of the pronoun in (2a) to a crucial distributional restriction on French lui, pointed out by Kayne (2001), stating that it is banned from structural Case positions, as witnessed by (9). However, as acknowledged by Reinhart and Reuland themselves, the Case- and-Chain theory does not suffice to account for the distribution of simplex and complex pronouns in all contexts; it is but one ingredient of their intricate theory of referential dependencies. For example, the interpretive contrast between (23a) (which forces the disjoint reading) and (23b) (which allows coreference) leads Reinhart and Reuland (1993) to distinguish predicative and non-predicative prepositions: in (23a), the preposition of is assumed to be a mere θ-assigner selected by the lexical head ashamed, whereas in (23b) the preposition behind is assumed to stand as a predicate head of its own – a P predicate – whose covert external argument needs to be controlled: in this particular example, it is said to be controlled by the event argument (the ‘looking’ event), so that the internal argument referring back to ‘John’ is realized as non-reflexive him with no Condition-B violation. Furthermore, the pronoun in (23a) must receive structural Case from the predicate ashamed, while the pronoun in (23b) must receive oblique Case from the preposition behind. Turning to English-French comparison, the Case-and-Chain theory leads us to assume that lui is specified for oblique Case in both (24a) and (24b), the French translations of (23a,b):

(24) a. Jeanz John

a have.prs.3sg

honte de shame of

luiz/k . 3m.sg

‘John is ashamed of him(self).’ b. Jeanz

John a have.prs.3sg

regarde look.pp

derrie`re behind

luiz/k . 3m.sg

‘John looked behind him.’

However, there may be some circularity in this description, for de seems selected by honte in French (24a) very much like of is selected by ashamed in English (23a). Ashamed is a denominal adjective whose complement is a former genitive replaced by [of+noun phrase] in Modern English (van Gelderen 1999, ex. [20]). Honte in French is a nominal whose “genitive” complement is similarly realized as [de+noun phrase]. Other evidence in support of a Case contrast between him


in (23a) and lui in (24a) is the fact that P-stranding is licensed in English but not in French (Reuland, c.p.; cf. Kayne 1981):

(25) a. Who is John ashamed of? b. *Qui

Who est-ce que +q

Jean John

a have.prs.3sg

honte shame

de? of

This contrast suggests that English proud+of (unlike French fier de) undergoes [PRED Pred+P] reanalysis and is thus capable of assigning structural Case to its complement. However, such examples as (26) indicate that P-stranding does not have to involve [PRED Pred+P] reanalysis:

(26) a. John wrote Mary his angry letter in the passageway. b. What part of the house did John write Mary his angry letter in ?

In (26b), the locative PP is not selected by the verb and the complement of in is unlikely to be Case-marked by the verb head. The acceptability of P-stranding thus does not provide a diagnostic test for structural Case assignment; hence we cannot infer from the contrast between (25a) and (25b) that the pronoun is specified for structural Case in English (23a) and for oblique Case in French (24a). Since Modern-English him is not specified for the accusative/oblique distinction in morphology, the main available evidence that English him and French lui are not similarly specified for syntactic Case in (23a) and (24a) is the interpretive contrast between him and lui in these examples – the very problem we are attempting to explain.

2.6. Soi as a blocking factor

Basing myself on English-French comparison, I addressed in Zribi-Hertz (2003) the issue of the linguistic change which leads from the intensive to the reflexive use of M-pronouns: the distribution of lui-me me in Modern French is motivated by semantic properties, while the distibution of himself in Modern English is – for a subset of its occurrences – motivated by syntax. The problem is to understand how the occurrence of M-pronouns comes to be triggered by a syntactic property, and why this development has not occurred in French.

Defining binding as a local and obligatory relation, I assumed that while so-called “reflexive anaphors” are bound by an argument (a [+θ] antecedent), so-called “pronominals” are bound by a non-argument, a [−θ ] operator – a discourse topic syntactically represented in the domain periphery (cf. Rizzi 1997):


(27) a. [cp[top øk] [tpJohn is {proud/jealous} of himk]]

b. [tp Johnz is {proud/jealous} of himself z]

Following Zribi-Hertz (1995) and Ko nig and Siemund (1999),16 I further assumed that simplex pronouns such as him or lui may a priori (i.e. on the basis of their φ-features) be bound by [−θ] or [+θ] antecedents, but that [+θ] binding is marked with respect to [−θ] binding. Evidence supporting this view is the behaviour of French lui, which may always be topic-bound regardless of lexical semantics, while its binding by an argument is crucially sensitive to lexical semantic features:

(28) a. [topøk] [tp Jeanz John

est be.prs.3sg

{fier/jaloux} de proud/jealous of

luik] 3m.sg

‘Johnz is {proud/jealous} of himk.’ b. [tp Jeanz

John est be.prs.3sg

{fier/*jaloux} proud/jealous

de luiz] of 3m.sg

‘John is {proud/*jealous} of himself.’

Intensified pronouns provide an optimal strategy for allowing the reflexive reading in [+od] contexts: within the lui-me me complex word, the destressed pronoun lui precludes the ostensive use, thus forcing an endophoric reading (Zribi- Hertz 1995); the intensive marker calls for the “most central” binder (in Ko nig’s sense) – the [+θ] antecedent (as opposed to the [-θ] topic) if no further context is provided. In the resulting distributional pattern, topic-binding is always available for lui, while argument-binding is always available for lui-me me; in other words, the topical (A-free) reading is the unmarked reading for lui (cf. [28]), while the reflexive (A-bound) reading is the unmarked reading for lui-me me:

(29) a. Jeanz est jaloux de luiz-meˆme. ‘John is jealous of himself.’

[+od predicate]

b. Jeanz est fier de luiz-meˆme. [−od predicate] ‘John is proud of himself.’

16 A similar but not quite identical view put forward by Levinson (1991) is that argument binding is a priori marked, regardless of predicate semantics; in other words, reflexive readings are as such marked, with respect to argument structure. As also emphasized by Ko¨nig and Vezzosi (2004), this generalization is proved to be too strong by such examples as French (2a,c): argument binding is semantically marked only with [+od] predicates.


The distribution of Modern English him and himself, which motivated Chom- sky’s (1981) Standard Binding Theory, may thus be viewed as a regularization of the state of affairs illustrated in (28)–(29):

(30) him and himself in a predication (“local”) domain

The English distributional pattern may be described a syntax-driven development of the grammar instantiated by Modern French: whereas the distribution of lui and lui-me me in (28)/(29) is sensitive to a lexical-semantic property (±od), the distribution of him and himself is motivated in (27)/(30) by the syntactic contrast between argument-binding and topic-binding. This development is consistent with the Chomskyan view of grammatical economy summarized by Reuland (1999), according to which language-processing mechanisms based on discrete categories such as [±θ ] are a priori more economical than mechanisms based on continuous categories such as other-orientation or centrality. In line with this general approach, I suggested in Zribi-Hertz (2003) that French lui and lui-me me should be expected to eventually undergo a “syntacticization” process leading to their distributional complementarity in examples such as (2).

However, this complementarity does not obtain in today’s French, which suggests that something in the grammar hinders the expected development. In Zribi-Hertz (2003), I proposed to link the present state of affairs to the special properties of the French pronoun soi, which has no counterpart in English. French soi is a non-clitic 3rd-person pronoun historically derived from Latin se. Like Latin se, and like its modern clitic counterpart se, soi is unspecified for gender and number. As a strong pronoun, soi is available in prepositional contexts. In French textbooks and dictionaries, soi is commonly labelled re fle´chi (‘reflexive’). In archaic Old French, soi could be bound by referential antecedents, but it very early competed with lui in such contexts. Some such occurrences of soi are still attested in modern literary texts, as witnessed by the three examples in

(31), drawn from Rey-Debove and Rey (1993: •••):17

17 In these and further similar examples adapted from attested written productions, the pronoun which occurs in the original text is boldfaced. All proposed translations are my own.

binder rθ tθ

pronoun himself

him


(31) a. Un

a.m.sg hommez man

droit, uptight

ferme , withdrawn

su r de sure of

{soi/lui}z . (Sartre) soi/ 3m.sg ‘An uptight, withdrawn man, sure of himself.’

b. Il 3m.sg

s’expliquait understand.impf.3sg

trop too

bien well

que that

le def.m.sg

comtez count

fu t be.sbjv.impf.3sg

a peine hardly

ma ıtre de master of

{soi/lui}z . (Bourget) soi/ 3m.sg ‘He readily understood why the count should be losing control over himself.’

c. Ellez 3f.sg

se moquait de not-care.impf.3sg of

s=a 3sg=f.sg

fille daughter

et and

ne neg

pensait think.impf.3sg

qu’ que

a about

{soi/elle}z . soi/ 3f.sg

‘She didn’t care about her daughter and thought only of herself.’

In Modern French grammars, however, soi is usually described as restricted to quantified antecedents, as in (32) ([32a,c,d] are quoted by Grevisse 1986: •••); in such cases, soi also competes with lui, except when the binder is arbitrary on [32a] or pro [32b]) – in this case lui is disallowed:

(32) a. Onz

one

ne neg

peint paint.prs.3sg

bien well

que que

{soi/*lui}z soi/3m.sg

et and

les def.pl

s=iens. (France) 3sg=pl.m ‘One can only properly paint oneself and one’s own.’

b. [øz] travailler work.inf

pour for

{soi/*lui}z

soi/3m.sg est be.prs.3sg

une a.f.sg

chose thing

bien very

agre´able. pleasant

‘To work for oneself is a very pleasant thing.’ c. Chacunz

everyone doit must

s’inte´resser a pay attention to

{soi/lui}z . (Beauvoir) soi/3m.sg

‘Everyone must pay attention to oneself.’


d. Chaque

every hommez man

renferme enclose.prs.3sg

en within

{soi/lui}z

soi/3m.sg un a.m.sg

monde world

a part. of its own

(Chateaubriand)

‘Every man encloses within himself a world of its own.’

For many Modern-French speakers, however, soi is unavailable in such cases as (32c,d) and may only be bound by on or arbitrary pro, as in (32a,b) (cf. Brandt 1944; Zdobyck 1998), or else occur as a free arbitrary pronoun, as in (33):

(33) Il it

est be.prs.3sg

clair clear

qu’ that

il y a there is

dans in

toute every

librairie bookshop

des pl

livres book.pl

inte´ressants interesting

pour for

soi. soi

‘Any bookshop is bound to contain some books of interest for oneself.’

As shown by Brandt (1944) and confirmed fifty years later by Zdobyck (1998), the most advanced dialectal varieties of French treat soi as a strong, non- nominative allomorph of on. My own assumption in Zribi-Hertz’s (2003), which is akin in spirit to Pica’s Avoid Pronoun approach (cf. Section 2.3), was that the availability of soi as an A-bound pronoun hinders the development of lui-me me as a reflexive marker in prepositional contexts. My prediction was therefore that the intensive > reflexive development of lui-me me should only occur in dialects whose grammar no longer licenses soi in such examples as (31) and (32c,d).

However, the blocking effect of soi on the evolution of lui-me me cannot suffice to explain the contrast between French and English addressed in this study. First, no observable evidence suggests that lui-me me is currently under- going the intensive > reflexive change in spoken French, even in dialects which sharply ban soi from (31) and (32c,d) as does my own baby-boom/Paris variety. The data in (34) below rather suggest that even in this dialectal variety of French, lui-me me is an intensified pronoun whose occurrence is motivated by focus, rather than by A-binding:

(34) [Speaker showing the hearer a photograph] a. Regarde

look comme how

chacunz

everyone ici here

est be.prs.3sg

fier de proud of

{˚ soi/lui}z !18

soi/3m.sg ‘Look how proud of themselves everyone is!’

18 Soi indicates that acceptability is restricted to archaic grammars of French.


lui

b. Regarde

look comme how

Jeanz John

est be.prs.3sg

fier proud

de luiz/k of 3m.sg

(??-meˆme}! int Look how proud of him(self) John is !’

c. Regarde look

comme how

Jeanz

Jean est be.prs.3sg

jaloux jealous

de ∗z/k of 3m.sg

(*-meˆme)! int ‘Look how jealous of him(self) John is!’

In the construction Regarde comme+Predication, the new information is con- veyed by the comme-clause. Hence, no constituent within this clause should fall under narrow focus. (34a) shows that the dialectal grammar under discussion fails to acknowledge soi as a reflexivity marker. lui-me me, however, sounds awkward not only in (34c) where the predicate is [+od], but also in (34b) with a [−od] predicate. This contraint, which does not carry over to himself in the En- glish translations, is expected under the assumption that French lui-me me must always be under narrow focus. It furthermore appears that the French-English contrast between simplex (him, lui) and complex (himself, lui-me me) pronouns reaches beyond the issue of reflexivity and bound anaphora, for French lui-me me also appears more restricted in its distribution than English himself in A-free contexts:

(35) a. That picture of her(self )z on the front page of the Times confirmed the allegations Maryz had been making over the years. [adapted from Pollard and Sag 1992: 264]

b. Cette dem.f.sg

photo picture

d’ellez of 3f.sg

(??-meˆme) a int on

la def.f.sg

Une front page

du of.def.sg

Monde Monde

a have.prs.3sg

confirme confirm.pp

ce that

que which

Mariez

Mary re´pe tait repeat.impf.3sg

depuis since

des pl

anne´es. year.pl

‘That picture of her(self) on the front page of Le Monde confirmed what Mary had been repeating for years.’

While the complex pronoun is acknowledged as well-formed in (35a) by all my English-speaking consultants, its French analogue in (35b) is felt as odd by all the French speakers I questioned. This suggests that some other factor must be


at work in the distributional contrast between English him(self) and French lui(me me). In what follows I will explore the assumption that one such factor is prosody.

3. The prosodic factor

The comparative study of English and French prosody is a vast issue that goes far beyond the present study (cf. Selkirk 1984). In what follows I would simply like to suggest that the different prosodic properties of English and French pronouns play a determining role in their semantic development. To present the prosodic properties which seem to me relevant, I will first lay out a few preliminary descriptive assumptions borrowed from Bu ring (1997, 2007), whose work on prosody and information structure encompasses many previous results. I will then consider the two major uses of himself (A-free, and A-bound) from a prosodic perspective, showing that three main prosodic patterns may be distinguished: (a) himself under primary accent (himself1 ); (b) himself under secondary accent (himself2); (c) deaccented himself (himself◦ ), which, I shall argue, may correlate with two types of information structure. I will then show that the (a) pattern is the only one which is available for French lui-me me, and will then proceed to derive this English-French contrast from the different prosodic properties of pronouns in the two languages: English pronouns are “leaners” (Zwicky 1982), while French non-clitic pronouns may be shown to be prosodically strong, a property which could correlate with the phrase-final accent characteristic of French. I will argue that the prosodic properties of pronouns contribute to explain why himself has a wider distribution in English than does lui-me me in French, and why himself, unlike lui-me me, has developed into a syntax-driven reflexive marker.

3.1. Some preliminaries

As pointed out by Selkirk (1984) and recalled by Bu ring (1997, 2007), in English and other Germanic languages, primary clausal accent signalling focus normally falls on the rightmost constituent inside the verb phrase, and the position of primary accent is independent of the scope of focus. Thus in the following examples, primary accent falls on the object Mary in the sentence He saw Mary both when focus has narrow scope on the object, as in (36a’), and when it has wide scope on the entire verb phrase, as in (36b’):

(36) a. Who did John see on Thursday night? a’. He saw Mary. [narrow focus on object]


b. What did John do on Thursday night? b’. He saw Mary. [broad focus on VP]

Bu ring calls integration the prosodic process which, in (36b’), merges the head and its complement to form a single prosodic unit: in this case, although primary accent falls on Mary, the V head is semantically integrated in the focused constituent saw Mary.

Another prosodic property which will be useful below is the deaccenting of given constituents. Bu ring emphasizes the fact that the deaccenting of object pronouns exemplified in (37c) echoes the deaccenting of lexical objects construed as given exemplified in (37b):

(37) Why does John keep criticizing Mary? a. — Because he is jealous of Mary. b. — Because he is jealous of her.

In (37a,b), the referent of the object is construed as given information; correlatively, focus has narrow scope over the predicate head jealous. In the French analogues of these examples, we note that pronouns contrast with lexical noun phrases with respect to prosody:

(38) Pourquoi est-ce que Jean passe son temps a critiquer Mariek? (= [37]) a. — Parce qu’il est jaloux de Marie. (= [37a]) b. — Parce qu’il enk est jaloux. (= [37b]) c. — Parce qu’il est jaloux d'ellek. (= [37b]) d. — *Parce qu’il est jaloux d’ellek.

In (38b), the discourse-given lexical complement of jaloux is deaccented, as in example (37b) in English. In (38c), the inherent prosodic deficiency of the clitic en echoes the deaccenting of of her in (37c). In (38d), however, the pronoun does not undergo deaccenting, even though it is discourse-given under the ‘k’ index: it is pronounced under primary accent together with the predicate head jaloux.

A third prosodic property relevant for what follows is the notion of secondary accent. Although primary accent regularly falls on the rightmost constituent in the verb phrase, other constituents in the sentence may bear secondary accents. In the following example, due to D. Bu ring (p.c.), the noun phrase dogs is accented although it is topical. Pitch diagrams based on recordings however show that the accent on the topic is secondary, i.e. less prominent than the focal accent on happy:


(39) Why did you buy a dog?

— Because dogs make me happy. 2 1

3.2. Eliciting the prosodic properties of himself and lui-me me

I will now proceed to show that himself may exhibit three different prosodic patterns: it may bear primary accent and be read as a narrow focus; it may bear a secondary accent; and it may be deaccented. I will illustrate each class of examples and will propose French translations along the way, in order to compare the distribution of French lui-me me to that of English himself.

3.2.1. himself under primary accent ( himself1)

As recalled above in Section 3.1, this prosodic pattern may a priori correspond to two types of information structure: narrow focus on the pronoun, or broad focus on the phrase which contains the pronoun. I shall consider each case separately.

himself1 may be A-free or A-bound.

A-free himself1 triggers a contrastive construal of its referent. A subclass of cases correspond to what Keenan (1988) called “complex anaphors”, which include conjunction and disjunction constructions, exemplified in (40a-e), and cases where contrast is signalled by an overt restrictive-focus particle such as only, as in (40f):

(40) a. Johnz believes that letter should be sent to both Mary and {him/

himself}z . a’. Jeanz pense que cette lettre devrait e tre envoyeé a la fois a Marie

et a luiz(-me^me). b. Johnz believes that letter should be sent to either Mary or {him/

himself}z . b’. Jeanz pense que cette lettre devrait e tre envoyeé soit a Marie, soit

a luiz (-me^me). c. Johnz believes that letter should be sent to everyone except {him/

himself}z . c’. Jeanzpense que cette lettre devrait e tre envoyeé a tout le monde

sauf a luiz(-me^me). d. Johnzthinks that Mary is in love with {him/himself}z, not Peter. d’. Jeanz pense que Marie est amoureuse de luiz (-me^me), pas de

Pierrez .


e. Johnz knew that people called Paul a crook, and that it could have been said also of {him/himself}z .

e’. Jeanzsavait que les gens traitaient Paul d’escroc, et qu’on aurait pu en dire autant de luiz(-me^me).

f. Johnzthinks that Mary hates only {him/himself}z . f’. Jeanz pense que Marie ne de´teste que luiz (-me^me).

Throughout (40), himself alternates in English with accented h ım under a non- reflexive reading. The simplex and complex pronouns both bear primary accent and may both be co-indexed with John. In the French analogues, lui and lui- me me similarly alternate under the ‘z’ index.

As rightly emphasized by Baker (1995), many of those occurrences of En- glish himself drawn from literary texts and which seem to violate the Binding Condition A (Zribi-Hertz 1989; Pollard and Sag 1992) are read as contrastive. This means that they should be pronounced under primary accent signalling narrow focus on the pronoun. Here as in (40) above, lui and lui-me me alternate in the French translations:19

(41) a. Hez [Zapp] sat down at the desk and opened the drawers. In the top right-hand one was an envelope addressed to {him/himself }z.

(Lodge) a’. Ilz s’assit au bureau et ouvrit les tiroirs. Dans celui du haut, a

droite, se trouvait une enveloppe adresse´e a luiz(-me^me). b. And that was exactly it, hez thought, he really didn’t care too much

what happened to {him/himself }z. (Highsmith) b’. C’e tait exactement c a, songea-t-ilz , ilz se fichait un peu de ce qui

pouvait luiz arriver a luiz (-me^me).20

himself1 may also be A-bound, as in (42). In such cases it does not alternate with simplex him. In the proposed French translations, lui(-me me) is available either as a clitic doubler (an adjunct), or as a P-governed argument:

(42) a. Hez sometimes felt that by [øk] torturing her*(self)k, hisz daugh- terk was torturing himz.

a’. Ilz sentait parfois qu’[øk] en sek torturant ellek (-me^me), sz=a fillek lez torturait luiz.

19 As in previous similar examples (cf. [31], [32]), the boldfaced form in each English example is the attested one, and the proposed French translation is my own.

20 In (41b’), English himself is translated in French by a pronoun-doubling construction (...lui arriver a lui[-meˆme]), since the reflexivized argument is a dative (see Section 1).


b. Johnz lit a cigarette for Mary, and then hez lit one for him*(self)z . b’. Jeanz alluma une cigarette pour Marie, puis ilz {en alluma une

pour/s’en alluma une a } luiz (−me^me). c. Since the boyz didn’t want it, wek decided [øk] to keep the model

for {∗us/ourselves}k . c’. Puisque l’enfant n’en voulait pas, nousz de cida mes de garder la

maquette pour nousz(−me^mes). d. Hez pulled the trigger first on his wife, then on himz∗(self). d’. Ilz tira d’abord sur sa femme, puis sur luiz (−meme).

Two important contrasts between French and English are revealed by these examples: (a) in French, whenever the internal argument is accusative or dative, it must be realized as a clitic; the strong pronoun is then licensed as an adjunct (cf. [41a’]); (b) lui is available alongside lui-me me throughout (42) – lui-me me is never obligatory.

Another class of cases is illustrated below by the two sets of examples in (43) and (44):

(43) a. Johnz eventually realized that Mary was {taller than himz (*self )/

taller than himz/taller than himselfz }. a’. Jeanz s’est finalement rendu compte que Marie e tait plus grande

que lui ( ??−me^me)z. b. Johnz thinks that grants should be given to linguists {like him

(*self )z/like himz/like himselfz}. b’. Jeanz pense que les bourses devraient e tre attribue´es a des lin-

guistes tels que lui (??−me^me)z. (44) a. Johnz put the book {behind him(*self )z/ behind himz/behind

himselfz }. a’. Jeanz a mis le livre {derrie,re lui (-me^me)z.}. b. Johnz pulled the cart {towards him(*self )z/towards himz /towards

himselfz }. b’. Jeanz a tire le caddy vers lui (-me^me)z.

Here as in (40)–(42), himself bears primary accent since it occupies the rightmost linear position within the predicate. And here as in (40)–(42), him alternates with himself1 under narrow focus. Unlike in (40)–(42), however, him in (43)–(44) may also be deaccented, with primary accent falling on the preceding lexical word. Unlike him, himself1 cannot undergo deaccenting in the contexts under discussion. Semantically, deaccented him correlates with either broad focus on the predicate or narrow focus on the accented head:


(45) What did Johnz find out about Mary?

a. — [That she was taller than {him(*self )/∗him/*him self}z].21 [broad focus on predicate phrase]

Mary was always convinced that she was shorter than John. b. — No. Johnz just found out that she is actually taller than

{him(*self )/∗him/*himself}z . [narrow focus on adjective]

Johnz must fear that Mary should be taller than hisz mum. c. — No, but hez fears she might be taller than {him/himself}z .

[narrow focus on pronoun]

(46) What did Johnz do with the book? a. — [Hez put it {behind him(*self )/ *behind him/ *himself}z .

[broad focus on predicate]

Did Johnz put the book behind {him/ (*self )/∗him/*himself}z ? b. — No, hez put it next to {him/*himself/*him/*himself}z .

[narrow focus on preposition] Did Johnz put the book behind Mary? c. — No, hez put it {*behind him(self )/behind {him/himself}z }.

[narrow focus on pronoun]

In the French translations of (45)–(46) proposed in (47) and (48), lui-me me only allows readings where it is construed under narrow focus, although even in such contexts simplex lui with primary accent remains optimal:

21 Acceptability judgements regarding pronouns in comparative constructions such as (43a)/ (44) involve a good deal of variation across speakers. English-speaking school children are taught that one must say (a) John is taller than I (am) and must discard (b) John is taller than {me/myself } as ungrammatical. It is however clear that most English speakers use (b) in informal speech. The debate regarding the choice between the simplex pronoun (me, him) or the complex self form is of another nature. Globally, British-English speakers are less reluctant than American-English speakers to accept himself in (45c)–(46c), a fact in keeping with Baker’s (1995) intuition that A-free himself is a British dialectal variant of discourse-linked contrastive him in American English.


(47) Qu’est-ce que

what +q Marie? Mary

Jeanz John

a might

pu have

de´couvrir discover(ed)

a propos de about

‘What might Johnz have discovered about Maryk?’ a. — [Qu’

that elle 3f.sg.nom

est be.prs.3sg

plus grande taller.f.sg

que than

luiz(∗−meme)]. 3m.sg int ‘That shek is taller than himz.’ [broad focus on predicate]

Marie Mary

a have.prs.3sg

toujours always

ete be.pp

convaincue convince.pp.f.sg

qu’ that

elle 3f.sg

e tait be.impf.3sg

plus petite shorter.f.sg

que than

Jean. John

‘Mary was always convincd that she was shorter than John.’ b. — Non.

no Jeanz John

s’est rendu compte realize.pst

qu’ that

ellek 3f.sg

est be.prs.3sg


que than

luiz (*-meˆme). 3m.sg int

‘Not always. Johnz realized shek is actually taller than himz.’ [narrow focus on adjective]

Jeanz Johnz sz=a hisz

doit must me,re? mum

craindre fear

que that

Mariek

Maryk

ne soit plus grande should be taller

que than

c. — Non, no

ilz

3m.sg.nom craint fear.prs.3sg

surtout qu’elle mostly that 3f.sg.nom

ne neg

soit be.sbjv.prs.3sg


que than

luiz(-me^me). 3m.sg int

‘No, he mostly fears she might be taller than {him/himself}z .’ [narrow focus on pronoun]


(48) Qu’est-ce que

what +q Jeanz John

a have.prs.3sg

fait do.pp

du of.def.m.sg

bouquin? book

‘What did John do with the book ?’ a. — Ilz

3m.sg.nom l’ 3sg.acc

a have.prs.3sg

pose put.pp

derrie,re behind

luiz(*-me^me). 3m.sg int ‘He put it behind him.’ [broad focus on VP]

Est-ce que +qu

Jeanz John

a have.prs.3sg

pose put.pp

le def.m.sg

livre book

derrie,re behind

luiz(*-meˆme)? 3m.sg int

‘Did John put the book behind him?’ b. — Non,

no ilz 3m.sg.nom

l’ 3sg.acc

a have.prs.3sg

mis put

a co te de next to

luiz (*-meˆme). 3m.sg int

‘No, hez put it next to himz.’ [narrow focus on P]

Est-ce que +qu Mariek ? Mary

Jeanz

John a have.prs.3sg

pose put

le def.m.sg

livrew

book derrie`re behind

‘Did Johnz put the bookw behind Maryk?’ c. — Non,

no ilz 3m.sg.nom

lw’ 3sg.acc

a have.prs.3sg

mis put.pp

derrie`re behind

luiz (-me^me). 3m.sg int

‘No, hez put itw behind {him/himself}z . [narrow focus on pronoun]


3.2.2. himself under secondary accent ( himself2)

In another subset of its occurences, himself bears a secondary accent within its clause. Correlatively, himself2 is not construed under narrow focus. As himself1, himself2 may be A-free, in which case it alternates with simplex him under a given referential index, or A-bound, in which case it does not. I shall consider each subclass of cases separately.

A-free himself2 is typically illustrated by its occurrences in picture nominals, whose interpretive properties have been much discussed in the linguistic literature (cf. Warshawsky 1965; Ross 1970; Cantrall 1974; Chomsky 1981; Kuno 1987; Pollard and Sag 1992; Reinhart and Reuland 1993). The English examples in (49) are adapted from Pollard and Sag (1992); in the English examples, secondary accent is signalled by an acute accent (e.g. himse´lf, p´ıcture), while small capitals indicate primary accent:

(49) a. That {p ıcture of him/picture of himse lf }z in the museum bothered Johnz.

a’. Ce portrait de luiz( ??-meˆme) au muse´e tracassait Jeanz. b. That {p ıcture of her/picture of herse lf }k on the front page of the

Times made Maryk’s claims seem quite ridiculous. b’. Cette photo d’ellek( ??-meˆme) a la Une du Monde rendait les

alle´gations de Mariek tout a fait ridicules. c. Johnz’s campaign requires that {p´ıctures of him/pictures of him-

se lf }z be placed all over town. c’. La campagne e´lectorale de Jeanz requiert que des photos de

luiz ( ??-meˆme) soient affiche´es dans toute la ville. d. Johnz’s intentionally misleading testimony was sufficient to ensure

that there would be {p´ıctures of him/pictures of himse lf }z all over the morning papers.

d’. Le te´moignage de libe re´ment mensonger de Jeanz devait suffire a faire appara ıtre des photos de luiz ( ??-meˆme) dans tous les journaux du matin.

In order to bring out the difference between primary and secondary accent on English himself, let us consider an ambiguous example such as (49c), repeated below in (50). Two different prosodic contours are available here, one for him- self1, as in (50a), and one for himself2, as in (50b) :

(50) For Johnz’s campaign I think we should put up some pictures of Bush all over town.


a. — This is quite unnecessary. On the other hand, Johnz’s cam-

paign requires that pictures of {him/himself}z be placed all over town.

What do we need to do to ensure John’s reelection? b. — First of all, Johnz’s campaign requires that {p´ıctures of

him/pictures of himse lf }z be placed all over town.

In (50a), the pronoun – him(self) – bears primary accent within the sentence and is read under narrow, contrastive, focus. In (50b), himself receives primary accent within its noun phrase domain (pictures of himse lf ). The crucial observa- tion here is that simplex him and complex himself contrast prosodically within picture noun phrases: simplex him is deaccented (> p ıctures of him), while complex himself is not (*p´ıctures of himself/pictures of himself). At clause-level, however, primary accent regularly falls on the predicate. Within the clause, himself therefore only receives secondary accent. In the French translations proposed in (49), complex lui-me me would only be felicitous under narrow focus, thus in such contexts as (50a), but not in (50b).

himself2 may also be A-bound. This typically happens when himself occupies a non-rightmost linear position within the predicate, e.g. when it fills the first object position in a double-complement constuction (cf. [51a,c]), the object position followed by a particle (cf. [51b]), or the so-called “ecm position” (cf. [51d]). In such examples, himself does not alternate with him under a given referential index. In the French translations of (51), lui(-me me) appears as an adjoined clitic doubler, since se is present on the verb; and simplex lui and complex lui-me me are equally infelicitous in such A-bound contexts: they are pragmatically unmotivated as restrictive-focus markers. This stands as a sharp contrast between French and English:

(51) a. Enjoying this moment of solitude, John poured himse lf a cup of tea.

a’. Savourant ce moment de solitude, Jean se versa ( ??a lui[-meˆme]) une tasse de the .

b. John propped himse´lf up on the bed with a couple of pillows. b’. Jean se cala ( ??lui[-meˆme]) sur le lit avec un ou deux oreillers. c. John congratulated himse lf on his decision to leave. c’. Jean se fe´licita( ??lui[-meˆme]) d’avoir pris la de´cision de partir. d. John used to consider himse´lf above romantic involvement. d’. Jean se conside´rait ( ??lui[-meˆme]) au-dessus des attachements

amoureux.


In the English examples in (51), himself does not undergo any radical deaccenting, nor any prosodic attachment to the verb. Its prosodic contour is the same as that of a non-presupposed lexical noun phrase in the same position:

(52) What did John do when the tray was brought to him? — He poured {himse lf/Ma ry} a cup of tea.

But although himself fails to be deaccented in (51), it only bears a secondary accent.

3.3. Deaccented himself (himself˚)

In a last class of cases, himself undergoes deaccenting. Such cases will be subdivided into two subsets.

In the first subset, the deaccenting undergone by himself boils down to the deaccenting of given information illustrated above in (37). As witnessed by the following examples, himself˚ may be A-free (53) or A-bound (54):

(53) a. a’.

b.

b’.

I hope Johnz didn’t realize that Mary is taller than himselfz . J’espere que Jeanz ne s’est pas rendu compte que Mariek est plus grande que luiz(*-meme). — No, but hez unfortunately did realize that she is smarter

than himself z. — Non, mais ilz a malheureusement bien vu qu’elle est plus

intelligente que luiz (*-meˆme).

(54) a. There are things I like about myself, and things I hate about myself.

a’.

b.

Il y a chez moi( ??-meˆme) des choses que j’aime, et d’autres que je de teste. John didn’t cut himself but he did burn himself.

b’. Jean ne s’est pas coupe mais c’est vrai qu’il s’est bru le (*lui- meˆme/ *lui-me^me).

Deaccenting of given information is available for lexical constituents in French, even if it involves disrupting the unmarked phrase-final stress pattern, as in (53b’). Deaccenting is however unavailable for lui-me me in French. We ob- serve that such complex pronouns are infelicitous in both (53) and (54). This is expected under the assumption that lui-me me is only licensed under narrow focus.

Another class of cases exemplified in (55):


(55) What’s up? Why are you looking so upset?

a. — John killed himself. b. — ?John killed {my dog/him}.

The deaccenting of the object in (55b) signals the referent of my dog or him as given, and therefore needs to be justified by a broader discourse context. Contrastively, the deaccenting of the object in (55a) does not necessarily correlate with this presupposition effect: under the indicated prosodic contour, (55a) may be read as a thetic clause, conveying nothing but new information, or as a predication associating a discourse-new predicate (killed himself ) with a discourse-given subject (John). The deaccented use of himself exemplified in (55a) seems characteristic of the unmarked, non-focused reflexive reading. Fur- ther English examples are given below in (56) with French translations, which show that unlike English himself, French lui(-me me) is infelicitous if it does not bear narrow focus:

(56) a. John walked to the front desk and introduced himself. a’. Jean se dirigea vers la re´ception et se pre senta (??lui[-me^me]). b. If John hadn’t burnt himself, he wouldn’t have screamed. b’. Si Jean ne s’e tait pas bru le ( ??lui[-me^me]), il n’aurait pas crie . c. When I walked in, John was facing the mirror, staring at himself. c’. Quand je suis entre , Jean e tait devant sa glace, en train de se

regarder ( ??lui[-me^me]) fixement. d. To hold a pork-stuffed bun in an overcrowded bus is a lousy idea,

John admitted to himself. d’. Jean se dit ( ??a lui[-meˆme]) que de tenir a la main une brioche

au porc au milieu d’un bus bonde e tait de´cide´ment une tre,s mauvaise ide e.

The position of primary accent in these examples may ambiguously trigger narrow focus on the verb (as in [54]), or broad focus on the predicate – the natural interpretation in (56) without further context. In this latter case, I propose to assume that himself is semantically integrated into the predicate, forming a reflexive predicate in the sense of Reinhart and Reuland (1993). This integration is signalled by prosodic reduction of the pronoun – Reflexive Deaccenting. A crucial contrast between English and French in (56) is that lui(-me me) cannot undergo Reflexive Deaccenting.


3.4. The impact of prosody on the semantic development of pronouns

The above description has brought out the two contrasts summarized below in (57) between English himself and French lui-me me:

(57) a. English himself is in complementary distribution with him in A-

bound contexts, while French lui-me^me alternates with lui under a given referential index, except in [-od] prepositional contexts (cf. [17]). In other words, himself stands as a syntax-driven reflexive marker (an A-bound “anaphor”) in a subset of its occurrences, while French lui-me^me is but an intensified variant of lui even when it is A-bound.

b. English himself may bear primary accent, secondary accent or be deaccented, while French lui-me^me must be under primary accent and construed semantically under narrow focus.

I now propose to derive these two properties from the following general contrast between English pronouns, and French non-clitic pronouns:

(58) a. English pronouns are prosodically weak, since they may undergo deaccenting and be realized as “leaners” (Zwicky 1982).

b. French non-clitic pronouns are prosodically strong, since they do not undergo deaccenting, unlike clitic pronouns, which inherently qualify as leaners (cf. Miller 1992).

English pronouns are less weak than French clitics, since they are not a priori closed to focal accent22,23They are on the other hand weaker than French non- clitic pronouns, since the latter do not undergo deaccenting. This is exemplified

22 Clitics may only be contrasted like sub-word-level morphemes or word parts, i.e. within their including constituents. Thus the acceptability of (i-b) in French is parallel to that of (i-a) in English: (i) a. Let’s not go to a motel, let’s go to a hotel. [Bolinger 1961]

b. Quand on rencontre un ministre femme il faut la saluer, pas le saluer. lit. ‘When you meet a female minister you must her greet rather than him greet.’

But clitics may not be construed under primary focus: (ii) Jean et Marie vont venir tous les deux demain. Tu comptes saluer qui?

‘John and Mary are both coming tomorrow. Who are you planning to greet?’ — *Je compte la saluer. lit. ‘I’m planning to her greet.’ (compare English: I’m planning to greet her.)


above in (38d), and below by the contrast between English (59) and its French counterpart in (60):. (59) How did the coachz feel when Paulk won the tournament?

— Hez felt very proud of him∗z/k . (60) Comment est-ce que l’entraˆıneurz a re´agi quand Paulk a gagne la

course? a. — Ilz s’est senti tre,s fier de luiz/k. b. — ??Ilz s’est senti tre s fier de luiz/k.

In (59), the pronoun undergoes “deaccenting of the given”, so that primary accent falls on the predicate head proud. In (60), on the other hand, the entire string tre s fier de lui is pronounced under high pitch corresponding to primary accent.24 The strength of French non-clitic pronouns could derive from more general prosodic properties of this language, which has no word-stress but only a phrase-final accent.

Let us now consider how these prosodic properties of pronouns might contribute to influence their interpretive properties. I argued above that although simplex personal pronouns such as him or lui are a priori open to both argument- binding and topic-binding, topic-binding stands as their unmarked interpretation, since it is insensitive to lexical semantic features. Consider the generalization proposed in (61):

23 A special case is the pronoun one, which (unlike other pronouns in the simplex HIM paradigm), can never bear stress (neither primary nor secondary). This leads to the contrasts illustrated below: (i) a. Johnz fears that some mischievous benefactor should send presents to

everyone except {him/himself}z. b. Onez sometimes fears that some mischievous benefactor should send

presents to everyone except {∗one/oneself}z. (ii) a. Like all feelings felt for {him/himself}z, John thought, it made himz sad.

b. Like all feelings felt for {∗one/oneself}z, Mrs Ramsay thought, it made onez sad. (adapted from Woolf , To the lighthouse)

(iii) a. Johnz thinks that pictures of {him/himself}z are pleasant to look at. b. Onez rarely thinks that pictures of {∗one/oneself}z are pleasant to

look at. Note that the prosodic weakness of one cannot be due to its semantic arbitrariness, since French arbitrary soi is regularly accented.

24 The details of the prosodic contour in (60a) are left here as an open issue. The only relevant element for this discussion is the fact that the pronoun in this example undergoes no prosodic reduction.


(61) All other things being equal, the deaccenting of him-type pronouns con-

tributes to favour their topic-binding over their argument-binding.25

This generalization is consistent with the idea that discourse-given information is deaccented (Ariel 1990; Bu ring 2007).This correlation is illustrated by French third-person pronouns, among which clitics (other than se26) must be topic-bound (cf. [16c], [20a], [21a]), while non-clitics are ambiguous between topic-binding and A-binding (cf. [2a], [20b], [21b]). Under the general assumption in (61), the fact that French non-clitics cannot undergo deaccenting might contribute to explain their persisting ability to be A-bound in Modern French, which in turn has a blocking effect on the development of lui-me me as a syntax- driven reflexive marker. As the intensified variant of a strong pronoun, French lui-me me may only be motivated by narrow focus on the pronoun, including in A-bound contexts such as (4). Independent empirical evidence supporting (61) is provided by the development of pronouns in French-based creoles: at an early stage of creole grammars, represented for instance by the most conservative (Northern) dialect of Haitian (cf. Zribi-Hretz and Glaude 2007), third-person pronouns (li in the singular) are ambiguously construed either as topic-bound or as A-bound in such examples as (62):

(62) Haitian Creole (Zribi-Hertz and Glaude 2007: •••)

Jan John

we see

l(i). 3sg

(i) ‘John saw him/her.’ (ii) ‘John saw himself.’

In more advanced varieties of Haitian, however, as well as in Martinican and Guadeloupean, for instance, (62), or its Martinique-Guadeloupe analogue, is only read as non-reflexive, and a bodypart possessive takes over the reflexive reading:

(63) Haitian Creole (Zribi-Hertz and Glaude 2007: •••)

25 An apparent counterexample to this generalization is one, mentioned in Note 22, which although always unstressed, is not topic-bound. This restriction may however be derived from the quantified (“arbitrary”) character of one, which a priori conflicts with topicality. In order to be topic-bound, pronouns must be made up of features allowing them to identify discourse referents.

26 Cf. Reinhart and Siloni (2005). French se partakes in a special morphosyntactic arity-reduction process internal to argument structure, and is thus immune to topic- binding.


Jan John

we see

te t head

li. 3sg

lit. ‘John saw his head’ = ‘John saw himself.’

This suggests that in the creoles under discussion, bodypart possessives are developing into syntax-driven reflexivity markers motivated by A-binding, an assumption confirmed by native speakers’ acceptability judgements: while some Haitian speakers straightforwardly discard the reflexive reading for (62), oth- ers who do accept it nevertheless remark that the bodypart construction in (63) would be optimal to convey this interpretation. Things are very different in French, where speakers unanimously favour, e.g., (2a) over (2c), under the reflexive reading. Interestingly, unlike the French non-clitic pronouns from which creole pronouns are historically derived, creole pronouns undergo a phonological reduction which leads to deaccenting: in Haitian (62), li drops its final vowel, as shown in (64a), while in Martinican/Guadeloupean it drops its initial conso- nant so that the remaining vowel [i] is realized as a glide, as shown in (64b); in either case, the pronoun loses its syllabic autonomy:

(64) Haitian Creole (Zribi-Hertz and Glaude 2007: •••) a. Jan we li. > Jan we -l.

Martinican/Guadeloupan Creole b. Jan vwè li. > Jan vwè-y.

John see 3sg ‘John saw him/her’.

The semantic contrast between French and creole 3rd-person pronouns is pre- dicted under the generalization in (61): once they undergo a phonological reduction leading to deaccenting, creole 3rd-person pronouns select topic-bound over argument-bound readings and thus develop a “Condition B” effect; correlatively, a marked strategy needs to be developed for argument-binding (reflexive readings). This evolution fails to affect non-clitic lui in French, since the pronouns in this paradigm do not undergo prosodic reduction.

The prosodic weakness of English him could thus contribute to account for its having become restricted to topic-binding and excluded from A-binding, in other words, for its having inspired Chomsky’s Binding Condition B. As argued by Levinson (1991), the disjoint-reference effect associated with him may account for the development of intensified himself as a reflexivity marker. The Reflexive Deaccenting of himself which occurs in such examples as (56) on the other hand correlates with the reduced argument structure of semantically reflexive predicates, which involves two different θ-roles linked to the same


referent. As argued above in Section 3.2.3, the prosodic reduction which affects himself in (56) is different from the case exemplified in (53)–(54), which is simply an instance of “deaccenting the given”, involving narrow focus on the preceding lexical head. The fact that himself may undergo deaccenting seems paradoxical since self-marking was initially motivated by intensification. As a result of prosodic reduction, himself is open to three types of prosody in Modern English: primary accent, secondary accent, no accent. French lui-me me, on the other hand, is restricted to primary accent and does not stand as a syntax-driven reflexivity marker, two correlated properties under the proposed analysis.

A final remark is in order regarding the compatibility of the above assumptions with the well-supported Phonology-Free-Syntax Principle (cf. Miller, Pul- lum and Zwicky 1997), which states that phonological properties should as a whole be invisible to syntax. Thus, no syntactic-agreement rule should ever be restricted in its application to, e.g., words beginning with a vowel. The assumption, proposed above, that the prosodic properties of pronouns could play a relevant role in the development of interpretive properties, hence of their distribution, could appear as a violation of the Phonology-Free-Syntax Prin- ciple. However, the prosodic properties which are relevant to the evolution of pronouns do not pertain to morphophonology, but to phrasal prosody, which crucially contributes to encode information structure. Hence, what the evolution of pronouns is ultimately sensitive to is not phonology, but information structure, signalled by prosody. Since the sensitivity of anaphora to information structure is a well-supported assumption (cf. Chomsky 1977; Reinhart 1983; Kuno 1987; Erteschik-Shir 1997, a.o.), the idea that phrasal prosody may be a determining factor in the interpretation and distribution of pronouns leads to no paradox.

4. Concluding remarks

The analysis proposed in Section 3.3 incorporates several intuitions put forward in the past linguistic works surveyed in Section 2: that the occurrence of English himself (unlike that of French lui-me me) is syntax-driven in a subset of cases; that the interpretive properties of English pronouns are very much, though not exactly, similar to those of French clitic pronouns, rather than to those of French non-clitic pronouns; and that French non-clitic pronouns form, so to speak, a “fourth type” of linguistic expressions with respect to reference relations. The relevance of the prosodic factor in the evolution of pronouns of course does not preclude that other independent factors should be at work, for instance Case properties, or economy principles such as “Avoid Pronoun”. From a typological perspective, the above analysis predicts that all other things being equal, the


prosodic weakness of him-type pronouns in any language should contribute to favour disjoint-reference effects.

References Ariel, Mira

1990 Accessing Noun Phrase Antecedents. London: Routledge. Baker, C. Lee

1995 Locally free reflexives, contrast, and discourse prominence in British English. Language 71: 63–101.

Bolinger, Dwight 1961 Contrastive accent and contrastive stress. Language 37.1: 83–96.

Bouchard, Denis 1984 On the Content of Empty Categories. Dordrecht: Foris.

Brandt, Gustaf 1944 La concurrence entre soi et lui, eux, elle(s). Etude de syntaxe his-

torique franc¸aise. Copenhagen: Munksgaard. Bu¨ring, Daniel

1997 The Meaning of Topic and Focus – the 59th St. Bridge Accent, New York/London: Routledge.

2007 Intonation, Semantics and Information Structure. In: Gillian Ramc- hand and Charles Reiss (eds), The Oxford Handbook of Linguistic Interfaces, ***–***. Oxford: Oxford University Press.

Cantrall, William 1974 Viewpoint, Reflexives, and the Nature of Noun Phrases. The Hague:

Mouton. Cardinaletti, Anna and Michael Starke

1999 The typology of structural deficiency: a case study of the three classes of pronouns. In: H. van Riemsdijk (ed.), Clitics in the languages of Europe, 145–233. Berlin: Mouton-De Gruyter.

Chomsky, Noam 1977 Essays on Form and Interpretation. NewYork: Elsevier North Holland

Publications. 1981 Lectures on Government and Binding. Dordrecht: Foris.

Edmonson, Jerold and Frans Plank 1978 Great expectations: an intensive self analysis. Linguistics and Philos-

ophy 2.3: 373–413. Emonds, Joseph

1985 A Unified Theory of Syntactic Categories. Dordrecht: Foris. Erteschik-Shir, Nomi

1997 The Dynamics of Focus Structure. Cambridge UK: Cambridge Uni- versity Press.

Faltz, Leonard 1985 Reflexivization: a Study in Universal Syntax. New York: Garland.


Farmer, Ann and Robert Harnish

1987 Communicative reference with pronouns. In: J. Verschueren and M. Bertucelli-Papi (eds.), The Pragmatic Perspective, 547–565. Amster- dam: John Benjamins.

Gelderen (van), Elly 1999 Bound pronouns and non-local anaphors: the case of Earlier English.

In Z. Frajzyngier and T. Curl (eds.) Reflexives: Forms and Functions, 187–225. Amsterdam/Philadelphia: John Benjamins.

Grevisse, Maurice 1986 Le bon usage. Gembloux: Duculot.

Kayne, Richard 1975 French Syntax: the Transformational Cycle. Cambridge MA: MIT

Press. 1981 On certain differences between French and English. Linguistic Inquiry

12: 349–371. 2000 Parameters and Universals. Oxford: Oxford University Press. 2001 A Note on Clitic Doubling in French. In Cinque, G. and G. Salvi (eds.),

Current Studies in Italian Syntax. Essays Offered to Lorenzo Renzi, 189–212. Amsterdam: North-Holland (also in Kayne 2000).

Keenan, Edward 1988 On semantics and the Binding Theory. In J. Hawkins (ed.), Explaining

Language Universals, 105–144. Oxford: Blackwell. 2002 Explaining the creation of reflexive pronouns in English. In Minkova,

D. and R. Stockwell (eds.), Studies in the History of English: A Mil- lenial Perspective, 325–355. Berlin/New York: Mouton-De Gruyter.

Ko¨nig, Ekkehard 1991 The Meaning of Focus Particles: A Comparative Perspective. London:

Routledge. Ko¨nig, Ekkehard and Peter Siemund

1999 Intensifiers and reflexives: a typological perspective. In Frajzyngier and Curl (eds.): 41–74.

2000a The development of complex reflexives and intensifiers in English. Diachronica XVII: 39–84.

2000b Locally-free self-forms, logophoricity and intensification in English. English Language and Linguistics 4.2: 183–204.

Ko¨nig, Ekkehard and Letizia Vezzosi 2004 The role of predicate meaning in the development of reflexivity. In

Bisang, W., N. Himmelman and B. Wiemer (eds.), What Makes Gram- maticalization? ALook from its Fringes and its Components, 213–244. Berlin:Mouton de Gruyter.

Kuno, Susumu 1987 Functional Syntax:Anaphora, Discourse, and Empathy. Chicago: Chi-

cago University Press.


Levinson, Stephen

1991 Pragmatic reduction of the binding conditions revisited. Journal of Linguistics 27: 107–161.

Miller, Philip 1992 Clitics and Constituents in Phrase-Structure Grammar. New York:

Garland. Miller, Philip and Paola Monachesi

2003 Les pronoms clitiques dans les langues romanes. In Godard, D. (ed.), Les langues romanes, 67–123. Paris: CNRS Editions.

Miller, Philip; Geoffrey Pullum and Arnold Zwicky 1997 The principle of Phonology-Free Syntax. Journal of Linguistics 33.1:

67–90. Pica, Pierre

1984 Liage et contigu ıte . In Milner, J.-C. (ed.), Recherches sur l’anaphore, 119-164. Paris: Universite Paris-7, ERA 642.

1986 De quelques implications the´oriques de l’e tude des relations a longue distance. In Ronat, M. and P. Pica (eds.), La grammaire modulaire, 187-210. Paris: Editions de Minuit.

Pollard, Carl and Ivan Sag 1992 Anaphors in English and the scope of the Binding Theory. Linguistic

Inquiry 23.2: 261–303. Postal, Paul

1969 On so-called “pronouns” in English. In Reibel, D. and S. Schane (eds.), Modern Studies in English: Readings in English Transformational Grammar, 201–224. New Jersey: Prentice Hall.

Reinhart, Tanya 1983 Anaphora and Semantic Interpretation. New York: Croom Helm.

Reinhart, Tanya and Eric Reuland 1993 Reflexivity Linguistic Inquiry 24.4: 657–720.

Reinhart, Tanya and Tal Siloni 2005 The lexicon-syntax parameter: reflexivization and other arity opera-

tions. Linguistic Inquiry 36.3: 389–436. Reuland, Eric

1999 The fine structure of grammar: anaphoric relations. In Frajzyngier, Z. and T. Curl (eds.), Reflexives: Forms and Functions, 1–40. Amster- dam/Philadelphia: John Benjamins.

2001 Primitives of binding. Linguistic Inquiry 32.3: 439–492. Reuland, Eric and Tanya Reinhart

1995 Pronouns, anaphors and Case. In Haider, H., S. Olsen and S. Vikner (eds.), Studies in Comparative Germanic Syntax, 241–268. Dordrecht: Kluwer.

Rey-Debove, Josette and Alain Rey 1993 Nouveau Petit Robert. Paris: Dictionnaires Le Robert.


Rizzi, Luigi

1997 The fine structure of the left periphery. In Haegeman, L. (ed.), Ele- ments of Grammar. Handbook of Generative Syntax, 281–337. Dor- drecht/Boston/London: Kluwer.

Ronat, Mitsou 1982 Une solution pour un apparent contre-exemple a la theórie du liage.

Lingvisticae Investigationes 6.1: 189–196. Ross, John

1970 On declarative sentences. In: Jacobs, R. and P. Rosenbaum (eds.), Readings in English Transformational Grammar, 222–272. Waltham MA: Ginn and Co.

Selkirk, Elizabeth 1984 Phonology and Syntax. Cambridge MA: MIT Press.

Siemund, Peter 2000 Intensifiers in English and German: a Comparison. London: Rout-

ledge. Warshawsky, Florence

1965 Reflexivization I and II. Indiana University Linguistics Club [reprinted in J. McCawley (ed.), Notes from the Linguistic Underground. Syntax and Semantics 7: 63–84. New York: Academic Press].

Zdobyck, Virginie 1998 Le pronom SOI. Unpublished Master’s thesis, Universite Paris-8.

Zribi-Hertz, Anne 1980 Core fe rence et pronoms re flećhis: notes sur le contraste lui/lui-meˆme

en français. Lingvisticae Investigationes 4.1: 131–179. 1989 Anaphor binding and narrative point of view: English reflexive pro-

nouns in sentence and discourse. Language 65.4: 695–727. 1995 Emphatic or reflexive? On the endophoric character of French lui-

me me and similar complex pronouns. Journal of Linguistics 31: 333– 374.

2003 Re flexivite et disjonction re´fe´rentielle en français et en anglais. In: Miller, P. and A. Zribi-Hertz (eds.), Essais sur la grammairecompareé du français et de l’anglais, 189–227. Saint-Denis: Presses Universi- taires de Vincennes.

Zribi-Hertz, Anne and Herby Glaude 2007 La re flexivite en ha¨ıtien: reéxamen et comparaison. In Gadelii, K. and

A. Zribi-Hertz (eds.), Grammairescreóles et grammaire comparative, 151–182. Saint-Denis: Presses Universitaires de Vincennes.

Zwicky, Arnold 1982 Stranded to and Phonological Phrasing in English. Linguistics 20: 3–

97.

from intensive to reflexive: the prosodic factor · from intensive to reflexive: the prosodic...

Documents